Machines can write songs and I have complicated feelings about it

The first time I heard a Suno-generated song that was genuinely good, not "good for AI" but actually good, I felt a strange mix of excitement and existential dread. Excitement because the creative possibilities are wild. Dread because I spent seven years learning guitar and this thing just... made a song. In twelve seconds.

AI music generation has improved faster than almost any other AI category. A year ago, AI music sounded obviously artificial. Tinny, repetitive, weirdly structured. Today, the best outputs are indistinguishable from amateur human productions, and occasionally they rival professional work. Whether that's thrilling or terrifying depends on whether you make music for a living.

I tested the four main AI music generation tools across four categories: full songs, background music for videos, specific genre requests, and commercial usability. Here's the honest rundown.

Suno: The obvious leader for full songs

Suno is the tool everyone talks about, and for once, the hype is justified. Give it a text prompt describing the song you want, and it generates a complete track with vocals, instruments, arrangement, and production. The quality is often startling.

What I tested

I asked for "an upbeat indie rock song about working late on a startup, male vocals, guitar and drums." The result: a genuinely catchy track with a proper verse-chorus structure, lyrics that actually made sense, a guitar tone that sounded organic, and vocals that were eerily convincing. Was it Radiohead? No. Was it better than a lot of music on Spotify? Uncomfortably, yes.

For genre versatility, I tried hip-hop, jazz, classical, electronic, and country. Suno handled all of them competently. The hip-hop and electronic tracks were the strongest. The jazz was decent but repetitive. The classical was the weakest, feeling more like "classical-flavoured background music" than actual composed classical music.

The quality question

Audio quality is solid. Not studio-master quality, but perfectly fine for YouTube, podcasts, social media, and most commercial uses. The vocals are the most impressive part. They sound human. Not perfectly human, but close enough that casual listeners won't notice.

Pricing and licensing

Free tier gives you 10 generations per day. The Pro plan at $10/month gives 500 generations and commercial usage rights. The Premier plan at $30/month gives 2,000 generations.

The commercial licensing is clear: Pro and Premier subscribers own the commercial rights to their generations. You can use them in videos, ads, and products. This is a big deal and something that not all competitors can match.

The verdict on Suno

If you want full songs with vocals, Suno is the clear winner. Nothing else comes close for complete track generation. The combination of quality, speed, and clear licensing makes it the default recommendation.

AIVA: The orchestral specialist

AIVA (Artificial Intelligence Virtual Artist) takes a completely different approach. It's designed for cinematic, orchestral, and emotional music. If Suno is a band, AIVA is an orchestra.

What I tested

I asked for "an epic cinematic score for a fantasy film trailer." AIVA delivered something that genuinely sounded like it belonged in a movie. Swelling strings, dramatic percussion, brass hits at the right moments. The composition showed understanding of musical tension and release that impressed me.

For a "calm ambient piano piece for a meditation app," AIVA produced something beautiful. Gentle, unhurried, emotionally resonant. This is where AIVA excels: it understands mood and atmosphere better than any other tool.

Where it falls short

AIVA doesn't do vocals. At all. If you need lyrics and singing, look elsewhere. It also struggles with modern genres. Ask for electronic dance music and you'll get something that sounds like a classical composer's interpretation of EDM, which is as awkward as it sounds.

The generation process is also slower than Suno. Where Suno produces a track in seconds, AIVA takes a few minutes, and the interface is more complex. You can choose instruments, adjust structure, and tweak arrangement, which is powerful if you know what you're doing and confusing if you don't.

Pricing and licensing

Free tier lets you generate up to 3 tracks per month, but you don't own them (AIVA retains copyright). Standard plan at EUR11/month gives you 15 tracks with licensing. Pro at EUR33/month gives unlimited generations with full copyright ownership.

That copyright distinction matters. On the free tier, AIVA literally owns the music, not you. You need to be on a paid plan to use the music commercially.

The verdict on AIVA

Best for cinematic, orchestral, and ambient music. If you make YouTube videos, films, games, or anything that needs a score rather than a song, AIVA is the specialist. Just don't ask it to write a pop song.

Soundraw: The customisable loop machine

Soundraw is the most practical tool if you need background music for content. It doesn't try to create complete songs. Instead, it generates customisable music tracks that you can edit section by section.

What I tested

I needed "upbeat corporate background music for a product demo video." Soundraw generated a track, then let me adjust the energy level, add and remove instruments, change the tempo, and rearrange sections. The drag-and-drop editor is intuitive, and within five minutes I had a track that perfectly matched the pacing of my video.

This is Soundraw's killer feature: the customisation. Every other tool gives you a finished track that you either like or don't. Soundraw gives you a starting point that you sculpt into exactly what you need.

Where it falls short

No vocals. The music, while customisable, is clearly loop-based. You can hear where patterns repeat, especially in longer tracks. It's perfectly fine for background music, but you'd never listen to a Soundraw track for pleasure.

The genre range is narrower than Suno or AIVA. It does corporate, chill, upbeat, and cinematic well. Ask for jazz, metal, or anything genre-specific and the results are generic.

Pricing and licensing

Plans start at $16.99/month. All paid plans include unlimited downloads and commercial licensing. No free tier, but there's a free trial.

The verdict on Soundraw

Best for content creators who need background music they can tweak to fit their videos exactly. The customisation is unmatched. Just don't expect it to create anything you'd add to a playlist.

Beatoven.ai: The mood-first approach

Beatoven.ai takes a mood-first approach. Instead of describing a genre or style, you describe the emotion you want the music to evoke: "hopeful," "tense," "playful," "melancholic." The AI composes a track designed to make listeners feel that specific way.

What I tested

I asked for a "hopeful" track for a charity video. The result was genuinely moving. Piano, subtle strings, a gentle build. It felt appropriate and emotionally on target.

For a "tense" track for a documentary about cybercrime, it produced something with low drones, staccato percussion, and an unsettling undercurrent that worked perfectly.

Where it falls short

Beatoven is limited in scope. It excels at short (1-3 minute) mood pieces for video content. Anything longer starts to feel repetitive. Anything outside the "background for video" use case feels like a stretch.

The audio quality is a step below Suno and AIVA. It's fine for background music, but you'd notice the difference on good speakers or headphones.

Pricing and licensing

Free tier gives 5 tracks per month. Pro at $6/month gives 25 tracks. Business at $20/month gives unlimited tracks. All paid plans include commercial use.

The verdict on Beatoven

The most affordable option for simple background music. If you make YouTube videos or podcasts and need mood-appropriate background music without overthinking it, Beatoven is efficient and cheap.

The copyright elephant in the room

Let's talk about what everyone's wondering: is AI music legal to use commercially?

The short answer: it depends on the tool and your plan. Suno, AIVA (on paid plans), Soundraw, and Beatoven all grant commercial licenses for music generated on their platforms. You can use these tracks in your videos, ads, and products.

The longer answer: the legal situation is genuinely uncertain. There are ongoing lawsuits about whether AI music models were trained on copyrighted music. If courts rule that the training data was used improperly, the licenses these companies grant might not hold up. For now, the tools themselves are clear that paid users have commercial rights, but "for now" is doing a lot of work in that sentence.

My practical advice: use AI music for content where the risk is manageable (YouTube videos, social media, internal presentations). For high-profile commercial use (TV ads, film scores, major brand campaigns), the legal uncertainty means you might still want human-composed music or properly licensed stock music until the courts sort this out.

Which one should you use?

Full songs with vocals: Suno. Nothing else is close. Cinematic and orchestral: AIVA. It understands musical emotion beautifully. Customisable background music: Soundraw. The editor makes all the difference. Quick mood-based tracks on a budget: Beatoven.ai. Simple, cheap, effective.

The fact that we're comparing AI music tools on quality rather than novelty tells you how far this has come. Two years ago, AI music was a curiosity. Today, it's a legitimate production tool. Whether that's wonderful or worrying probably depends on what you do for a living.